Abstracts

Rationale: Lennox-Gastaut Syndrome (LGS) is an epileptic encephalopathy associated with intractable seizures and cognitive impairment. We have recently shown that interictal discharges of LGS recruit diffuse areas of association cortex (Neurology, 2013;81(7):665-673), including elements of two cognitive networks that normally show anti-correlated signal fluctuations-the default mode network (DMN), which displays increased activity when the brain is at rest, and the task-positive network (TPN), which engages during tasks demanding attention. Here we sought to compare the interactions between these networks in a group of LGS patients and a group of healthy controls using resting-state functional MRI (rs-fMRI) and seeded functional connectivity. Methods: 17 patients (mean±1SD age 34.41±10.7 yrs) with LGS and 17 healthy controls (27.8±6.4 yrs) were recruited. Each subject underwent up to 60 mins of task-free fMRI in a 3T GE scanner. Standard image pre-processing included spatial normalisation, regression of nuisance covariates from white matter, cerebrospinal fluid, and global signal, and motion correction. For each subject, the mean time series was extracted from a seed in the right posterior cingulate (PC, involved in DMN), and then correlated with all brain voxels to produce individual maps of voxels positively and negatively correlated with the PC. Each map was normalised using Fischer's r-to-z-transformation and then combined across subjects in i) two-tailed, one-sample t-tests for both LGS and control groups separately to find regions of significant non-zero correlation; and ii) a two-tailed, unpaired t-test to examine significant differences in correlations between groups. Results: i) Controls: robust positive correlations were observed in other previously reported nodes of the DMN, including ventral medial prefrontal cortex (vMPC), angular gyri (AG), and superior frontal cortex (SFC), while negative correlations were seen in regions of the TPN, including intraparietal sulcus (IPS), anterior cingulate (AC), precentral cortex, insula, and middle temporal regions (MT)(Fig. 1a). ii) LGS: positive connectivity was preserved in some DMN regions, including AG and SFC, while absent in others, including vMPC. Negative correlations were seen in precentral cortex and insula, while absent in ACC, IPS, and MT (Fig. 1b). iii) The between-groups comparison (LGS>controls) confirmed that LGS patients had significantly reduced positive correlation with vMPC, and reduced negative correlation with bilateral IPS and left MT (Fig. 1c). Conclusions: LGS shows a breakdown of intrinsic functional network organisation both within the DMN and between the DMN and TPN. Altered interactions in these key cognitive brain networks may contribute to patients' cognitive impairment. These results support our previous observations that epileptiform activity in LGS is expressed through widespread areas of association cortex. We conceptualise LGS as a ‘secondary network epilepsy', where the electroclinical manifestations reflect an abnormal mode of cognitive network behaviour.